Microstructure-informed effective phase-field model for structural fracture of platelet-reinforced hyperelastic composites

Microstructure-resolved phase-field models capture crack–reinforcement interactions in hyperelastic composites, but repeated platelet-scale fracture calculations are costly for structural analysis. This work develops a microstructure-informed effective phase-field model for hexagonal boron nitride (hBN)-reinforced silicone rubber. Fractography supports a predominantly matrix-dominated, reinforcement-deflected fracture idealization and motivates a mechanical-role-separated closure. A finite-strain phase-field branch represents the fracture-active response, while a low-order auxiliary branch represents reinforcement-mediated load transfer. Both contribute to the effective stress and share one phase-field state, but only the tensile energy of the fracture-active branch drives fracture evolution. Four closure parameters are constrained using 33 proportional multiaxial loading paths and the complete pre-peak histories of all four in-plane first Piola–Kirchhoff stress components. A coefficient-separated reconstruction accelerates the offline search, while the spatial phase-field formulation is retained for structural calculations. The identified parameters are frozen and transferred to re-entrant-corner and edge-defect problems and to a separate five-repeat-per-formulation tensile dataset. Eleven of the 12 structural comparisons have pre-peak NRMSE below 3%, with a maximum of 3.732%, and retain the dominant localization patterns. For the 11 cases with complete terminal reference responses, fracture-displacement differences remain within approximately 3%. Experimental peak-normalized pre-peak RMSE ranges from 1.20% to 2.98%. Structural calculations are typically accelerated by about 4–7×, reaching 11.04× in the fastest benchmark. The framework therefore provides a compact, fully offline route to structure-level fracture prediction without online microstructural solves.